For Gramazeka (HO2, AO2, BRH-0, EBJ2)

[Norm]

I think you may be using a cruder TC calculation

I have a discussion on TC calculation at: https://www.qfit.com/CalculatingTrueCounts.htm

Actually, it's a bit more complex in CVCX/CVData. The CV code for TC calcs and deck estimates is in the neighborhood of 800 lines, not counting the user interface. Longer in V6 due to the addition of decimal insurance indices. Comparing systems can be difficult as the authors used different assumptions in calculating indices, and rarely state those assumptions. For strategy comparison, I never use exact cards for deck estimate as few people play that way. I usually set CV to recalc TC before insurance and after every card is drawn. You also need to decide whether the deck estimate uses the cards in the discard tray or includes cards on the table that have been seen.

[Cacarulo]

I have a discussion on TC calculation at: https://www.qfit.com/CalculatingTrueCounts.htm

For strategy comparison, I never use exact cards for deck estimate as few people play that way. I usually set CV to recalc TC before insurance and after every card is drawn. You also need to decide whether the deck estimate uses the cards in the discard tray or includes cards on the table that have been seen.

I agree, but changing the way of calculating the TC will not change the order. Probably using a different estimation of remaining decks, say to the nearest half instead of exact cards, the SCORE will be a bit lower, but the ranking will not be altered as long as we use an exact betting scheme.

Cac

[Norm]

Agree unless two strategies are extremely close -- in which case it doesn't matter.

The other problem is how many and which indices to use. I use the authors' indices. But, that doesn't tell you the relative value of the tags. For that matter, some authors use RA indices and most don't. Some authors also revise indices. Stanford has done this. Bryce used to make minor AOII adjustments now and again and email them to me.

[BJGenius007]

Theoretical Kelly bets for a 20000$ bankroll

Halves....44.43$....123.66$....200.92$....283.82$
EBJII......48.03$....123.48$....200.61$....286.53$
RPC.......54.05$....132.37$....218.93$....316.18$
Highlow..48.40$....127.02$....204.43$....293.89$

Practical Kelly bets for our 'green' player

Halves....50$....125$....200$....300$
EBJII.......50$....125$....200$....300$
RPC.......50$....125$....225$....325$
Highlow..50$....125$....200$....300$

Frequencies of the different TCs

System--TC1----TC2----TC3-----TC4----TC5 >--Total

Halves..11.64..6.70...3.86....2.67...2.95...27.82
EBJII....11.67..6.62...3.76....2.18...2.75...26.98
RPC.....11.78..6.50...3.59....2.02...2.37...26.26
Highlow11.81..6.50...3.69....2.10...2.57...26.67

That's all folks!



System.....c-Score

Halves.......32.33
EBJ II.........31.39
RPC..........30.89
Highlow.....29.76

Why using $20,000 bankroll and show c-SCORE?

Does c-SCORE inflate SCORE by 100% because SCORE is based on $10,000 bankroll. When you double your bet, your return is also doubled. While c-SCORE is in the range of 30 to 32, SCORE is only 15 to 16.

[Cacarulo]

Agree unless two strategies are extremely close -- in which case it doesn't matter.

The other problem is how many and which indices to use. I use the authors' indices. But, that doesn't tell you the relative value of the tags. For that matter, some authors use RA indices and most don't. Some authors also revise indices. Stanford has done this. Bryce used to make minor AOII adjustments now and again and email them to me.

That is true, for the comparison to be fair, we must use the same number of indices. Here one can choose between EV-maximizing indices or Risk Averse indices. In my comparisons I used EM-indices. It would not be fair to compare EM with RA.
I think that 22 indexes is more than enough but whoever wants to use more, just has to generate them and run the sims.
R22 (Revisited Catch-22) is an interesting set where the only thing I changed is 10vT for TTv4.
Actually 10vT as an EM-index has the peculiarity of worsening the SCORE, that's why I don't use it. Last but not least, for a comparison of SCOREs it is necessary to minimize the standard error. For that it is important to run about 50 billion rounds.

Sincerely,
Cac

[Gramazeka]

Why using $20,000 bankroll and show c-SCORE?

Does c-SCORE inflate SCORE by 100% because SCORE is based on $10,000 bankroll. When you double your bet, your return is also doubled. While c-SCORE is in the range of 30 to 32, SCORE is only 15 to 16.

c-SCORE systems published by Richard Raid in his book. (maybe)

Bankroll 20.000 and the rest of the data is already by Zenfighter.

[Cacarulo]

Nor would I take the indices proposed by the authors as gospel. It is best to generate them yourself.

[Gramazeka]

Nor would I take the indices proposed by the authors as gospel. It is best to generate them yourself.

The fact of the matter is that the system ratings of many reputable researchers do not match.

[Cacarulo]

The fact of the matter is that the system ratings of many reputable researchers do not match.

But it is up to you not to stay with the intrigue. What is the correct answer if there seems to be no general agreement? You have the tools and the procedure in your hands to clear up the doubt.

Sincerely,
Cac

[iCountNTrack]

I am joining a bit late to the discussion, but in order to have the most meaningful comparison of different counting systems everything needs to be standardized and that includes the simulator, the parameters used to generate the indices, and of course use a standardized metric to compare performance SCORE, N0 etc

[Cacarulo]

I am joining a bit late to the discussion, but in order to have the most meaningful comparison of different counting systems everything needs to be standardized and that includes the simulator, the parameters used to generate the indices, and of course use a standardized metric to compare performance SCORE, N0 etc

Exactly. That is why I made it clear which were the parameters used in the comparisons:

1) 6D,S17,DOA,DAS,SPA1,SPL3,NS,4.5/6
2) Counting system A and counting system B
3) R22 plays (C22 with TTv4 instead of 10vT). They can be more as long as the same amount and the same plays are used in both systems.
4) Estimation of remaining decks can be to the exact card or to the nearest half.
5) Floored and EV-Maximizing (No RA) indices - (Full deck TC)
6) No burn card
7) 50 billion rounds each
8) SCORE or N0 (Optimal betting ramp and EXACT bets)
9) 1-12 and 1-16 spread

I think that's all. Anyone can use a simulator like CVDATA/CVCX or SBA or some personal one and solve the mystery.

Sincerely,
Cac

[BJGenius007]

Exactly. That is why I made it clear which were the parameters used in the comparisons:

1) 6D,S17,DOA,DAS,SPA1,SPL3,NS,4.5/6
2) Counting system A and counting system B
3) R22 plays (C22 with TTv4 instead of 10vT). They can be more as long as the same amount and the same plays are used in both systems.
4) Estimation of remaining decks can be to the exact card or to the nearest half.
5) Floored and EV-Maximizing (No RA) indices - (Full deck TC)
6) No burn card
7) 50 billion rounds each
8) SCORE or N0 (Optimal betting ramp and EXACT bets)
9) 1-12 and 1-16 spread

I think that's all. Anyone can use a simulator like CVDATA/CVCX or SBA or some personal one and solve the mystery.

Sincerely,
Cac

4.5/6 penetration seems low. I only play when penetration is 5/6 or 5.5/6. Also isn't SCORE assuming the spread is 20 to 1?

[DSchles]

4.5/6 penetration seems low. I only play when penetration is 5/6 or 5.5/6. Also isn't SCORE assuming the spread is 20 to 1?

And you live in which country??!!

SCORE can assume any spread you like. Different SCOREs for different spreads. In BJA3, for 6-deck, we used 1-8, 1-12, and 1-16.

Don